Evaluation of Speech-Based Protocol for Detection of Early-Stage Dementia

摘要

This paper describes a study of a protocol and a system for automatic detection and status tracking of early-stage dementia and Mild Cognitive Impairment (MCI), from speech and voice recordings. The research has been performed in the scope of the EU FP7 Dem@Care project. We describe the speech and voice recording protocol, different families of vocal features as derived from the recorded data, the statistical properties of the vocal features, a classifier based on support vector machine, and the classification results. The vocal features we used detect the manifestation of dementia in the human voice and speech, in three axes: the impact of cognitive deficit and slower brain processing, the impact of certain mood states often observed in dementia, and the impact of impairments of the neuromuscular mechanism of the speech production. Our analysis is based on recordings of over 80 diagnosed subjects; it yields dementia and MCI detection equal-error-rate below 20%, and demonstrates the high value of using speech and voice analysis for automatic screening and status tracking of dementia from the very early stage of MCI.